From Centralized Management of Robot Swarms to Decentralized Scheduling

Abstract

Future IoT systems might be comprised of hundreds or even thousands of nodes. Managing all of these nodes should be done automatically, as individual programming is tedious and error prone. For this reason, we introduced the swarm scheduler in previous work. It receives a specification of the job to be done, and schedules it on the best candidate robot. While this scheduler works fine for a limited number of nodes, it is based on a centralized approach and will not scale well when the number of nodes reaches hundreds or even thousands. In this paper we discuss the advantages and disadvantages of the centralized approach and present two different decentralized versions we are currently investigating as well as some preliminary results.